A user requires different information according to the user and the user's state. Hereinafter, a context means a state of the user. When the context changes, the information required by the user changes. As a technology with which in a state in which a context changes, the information desired by the users is presumed by collecting and analyzing a record of a user behavior and the information desired by the user is provided based on a result of the presumption, the following technologies are used.
An information presentation apparatus described in patent document 1 includes a map information accumulation means, a POI (Point Of Interest) information accumulation means, a map information acquisition means, a POI information selection log management means, a POI information weighting means, and a screen output means. Operation of the information presentation apparatus will be schematically described below.
The POI information accumulation means accumulate POI information for showing stores and the like on a map. The map information acquisition means extract map information and the corresponding POI information based on a user's current position or a user's status. The POI information selection log management means manage a log of the POI information selected by the user in past times. The POI information weighting means calculate posterior probability with which each piece of POI information is selected by using a probability network model for selecting the POI information according to the user's status (season, time, and place) and perform weighting of each piece of POI information according to the posterior probability. The screen output means display the POI information ranked according to the weighting in a screen.
A content delivery system which included an in-vehicle apparatus (a content reception reproduction apparatus), a content server for delivering content, and a service server for transmitting content candidate list information which can be delivered by the content server to the in-vehicle apparatus is described in patent document 2. The in-vehicle apparatus described in patent document 2 presumes a content desired by a user, creates a time table including information on content to be delivered by the content server and a delivery time of the content, and transmits the created time table to the content server. A program for scheduling program which creates the time table in the in-vehicle apparatus selects the content. Operation of the selection will be schematically described below.
First, the program for scheduling program generates a Bayesian network with a random variable about content information (information on content, consideration, data capacity, and the like), a random variable about user information (sex, favorite genre, and the like), and a random variable about status (information on a car and information about congestion). The program for scheduling program inputs information about a content to the Bayesian network and calculates a probability distribution about the user information and the status information. The program for scheduling program determines a degree of equivalence between a result of this calculation and various information about the user and the status at the time of recommendation, and determines the content with a high degree of equivalence as the content suitable for the delivery. The content is delivered in decreasing order of the degree of equivalence.
A content providing apparatus described in patent document 3 includes a user information management unit, a behavior log management unit, a rule management unit, a status recognition unit, a process selection unit, and a content delivery unit. Operation of the content providing apparatus will be schematically described below.
When the status recognition unit of the content providing apparatus described in patent document 3 acquires input data in relation to a certain user from various information devices which can communicate with the content providing apparatus, the status recognition unit compares the input data with an information recognition rule held by the rule management unit and recognizes the user's status. When the status recognition unit acquires input data which is user information data and behavior log data that are held by a user status management unit and the behavior log management unit respectively, the status recognition unit performs the status recognition to the input data. The information recognition unit outputs a result obtained by performing the status recognition as status recognition data.
Next, the process selection unit compares information recognition data with a process selection rule held by the rule management unit and determines a type of content to be provided to the user. The process selection unit outputs a content creation request data based on the determined content type.
Based on the content creation request data outputted by the process selection unit, the content delivery unit selects, from a plurality of content creation means for creating the content, a destination to which the request is sent, and makes the destination create the content. The content delivery unit delivers the created content to a user terminal associated with the user.
An information providing system described in patent document 4 includes an information providing server and a terminal that are configured so as to be communicable with each other.
The information providing server includes a recommendation data creation unit and a recommendation data transmission unit. The recommendation data creation unit of the information providing server creates pieces of content that fits the terminal or is a favorite of the user based on the state of the settings of the terminal and content utilization frequency data of the user in the terminal that are received from the terminal. The recommendation data transmission unit transmits the pieces of the content, which are created, to the terminal.
The terminal includes a terminal state detection unit, a setting data transmission unit, a recommendation data reception unit, a frequency data update unit, a frequency data transmission unit, an order calculation unit, and a display process unit. The terminal transmits the data of the state of the settings of the terminal and the data of the utilization frequency of content to the information providing server and receives the recommendation data from the information providing server. The terminal calculates a priority order by reflecting not only a degree of favorite but also a terminal's state when content is utilized in past times, arranges the pieces of the content recommended based on the calculated priority order, and displays them.
A method in which, by excluding a context parameter to which it is determined that it does not affect value judgment criterion, the context parameter which affects a whole information selection is optimized is disclosed in non-patent document 1. In the method disclosed in non-patent document 1, a degree of influence on the value judgment criterion when a certain context parameter changes is evaluated by calculating an amount of change, when one context parameter is varied, of a discrimination plane of a user preference model constructed by the SVM (Support Vector Machine) by using the Monte Carlo method.